[Ffmpeg-devel] [bug] raw.c link failure on ff_ac3_parse_header

Justin Ruggles justinruggles
Sat Apr 14 19:01:34 CEST 2007


Diego Biurrun wrote:
> On Thu, Apr 12, 2007 at 11:36:27AM +0200, Michael Niedermayer wrote:
> 
>>On Thu, Apr 12, 2007 at 12:58:42AM +0200, Diego Biurrun wrote:
>>
>>>On Wed, Apr 11, 2007 at 01:10:06PM +0200, Michael Niedermayer wrote:
>>>
>>>>On Wed, Apr 11, 2007 at 11:22:51AM +0200, Diego Biurrun wrote:
>>>>
>>>>>On Tue, Apr 10, 2007 at 09:53:18AM +0100, M?ns Rullg?rd wrote:
>>>>>
>>>>>>Diego Biurrun <diego at biurrun.de> writes:
>>>>>>
>>>>>>>I'm thinking that it might be cleaner to move ff_ac3_parse_header out of
>>>>>>>ac3.c and into parser.c where the other parse functions are.  This would
>>>>>>>save some #ifdefs in ac3.c.
>>>>>>
>>>>>>I'm undecided on the location of these things.  In a way I like to
>>>>>>keep everything relating to a particular codec together, especially
>>>>>>when the decoder uses the same functions.  OTOH keeping the parser
>>>>>>code in one place has its virtues.
>>>>>
>>>>>Here's an alternative patch that moves the code.  It's slightly more
>>>>>intrusive since it has to move code around, but it adds only one #ifdef.
>>>>>
>>>>>I think I prefer this version.
>>>>
>>>>iam against moving ac3 related code from ac3 related files to parser.c
>>>
>>>But parser.c already contains code related to several other codecs like
>>>MPEG-4 and AAC, MP2, MP3, AC3.  The rest of the AC3 parsing code is
>>>there.  Also, without even more #ifdefs, when you need just this one
>>>function, you have to link all of ac3.o into libavcodec.
>>>
>>>The first patch adds 4 #ifdef conditionals, the second just one and to a
>>>file that already has quite a few.  My vote goes to this variant.
>>
>>my vote goes to putting all the parsing code in <codec>_parser.c
> 
> 
> IMO this is something that can be done later on.  In the meantime I
> think my second patch is the cleaner solution of the two.

I agree here.  I do like Michael's idea of separate files for different
parsers, but for now I agree with Diego that the function could be moved
out of ac3.c to fix the dependancy issue.

-Justin




More information about the ffmpeg-devel mailing list